A comparative analysis of merging strategies for satellite precipitation estimates and ground observations over Chinese mainland

Abstract

Satellite precipitation estimates are attractive for providing spatially continuous precipitation measurement on the global scale, but improvements in accuracy are still needed relative to the point-scale gauge observations. In this study, two integration frameworks named Geographical Differential Analysis (GDA) and Geographical Ratio Analysis (GRA) combined with two interpolation techniques named Ordinary Kriging (OK) and Inverse Distance Weighting (IDW) were adopted to merge the satellite precipitation with ground-based observations. The Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) research product was merged with the ground rain gauges derived from CGDPA (China Gauge-based Daily Precipitation Analysis) at a monthly scale over the Chinese mainland from 2011 to 2015, and the merged precipitation products were evaluated by 10-fold cross-validation. Results showed that the merged precipitation products based on the GDA integration framework had significant improvements relative to the original TMPA. Moreover, the method based on GDA combined with Ordinary Kriging (GDA-OK) delivered better performances compared to the GDA-IDW (method based on GDA combined with Inverse Distance Weighting) since the OK considers the global spatial correlation. However, the GRA-based merged precipitation products had systematic underestimations over Chinese mainland during all seasons and were not recommended for satellite-gauge precipitation merging over the large spatial scale of Chinese mainland. In addition, we found that the OK is quite sensitive to the density gauge network, the shrinkage of the number of rain gauges may result in great uncertainties and errors in the OK interpolation technique.